Research On The Multi-objective Intelligent Optimization Algorithm For The Static Virtual Topology Design In Three Layers Optical Network

With the rapid growth of communications traffic, the electronic bottleneck caused by the signal processing speed in the electronic domain is particularly prominent. Therefore, the research and application on wavelengths routed for Optical Transport Network (OTN) and Dense Wavelength Division Multiplexing (DWDM) become very important issues. Considering the fact that the practical application of SDH networks and technologies are still widely used, this dissertation studies the multi-objective optimization algorithm for the static traffic virtual topology design in SDH/OTN/DWDM optical transport networks.Sponsored by the project from the State High Technology Research and Development Project (863Project), the Key Technology Research and Experiment System based on PCE in Multi-layer Multi-region Optical Networks, the research foundation of the static traffic routing and resource optimization is laid briefly in this dissertation. In order to solve the problem of the static traffic virtual topology design, a multi-objective optimization model which includes the parameters, objective functions and constraint conditions is built for the practical SDH/OTN/DWDM optical transport networks. Then, a genetic algorithm based on the Pareoto optimality of consumption, delay and congestion (abbreviated as GA-POCDC) for logical topology design is proposed, and its crucial elements are studied. The proposed algorithm effectively combines the advantages of the parallel search ability of Genetic Algorithms and elitist strategy which could improve the quality of the evolutionary population quickly.The performance evaluation software of the proposed algorithm is implemented via C programming language. Under NSFnet, the comparison of simulation results of the proposed algorithm with those obtained from some exiting algorithms shows that GA-POCDC could find much better solution sets of uniform spread and convergence closer to the true Pareto-optimal front. The resultant set of near-Pareto-optimal solutions contains a number of nondominated solutions, and thus the user can judge relatively and pick up the most promising one according to the application specific requirements. The performance of the proposed algorithm has been validated on different network topologies. Finally, some aspects for future research are given after the summarization of the obtained achievements.